I would like to reconsider the flexible hierarchization framework we developed last time
instead, I would like to propose to explicitly bind parameter(s) to layers, each parameter belongs to a layer, each layer may have multiple parameters (1-n)
let's say we have a simulation function
def sim(x, y, z):
...
Then I could write
xp = Experiment(sim, inputgen={'x': ...})
to bind x and leave y and z unbound for upper layers
the package automatically detects that only x is bound
it also inspects the generator and if it takes additionally arguments such as y or z, it automagically supplies the values to it at runtime
I would like to reconsider the flexible hierarchization framework we developed last time
instead, I would like to propose to explicitly bind parameter(s) to layers, each parameter belongs to a layer, each layer may have multiple parameters (1-n)
let's say we have a simulation function
Then I could write
to bind x and leave y and z unbound for upper layers
the package automatically detects that only x is bound
it also inspects the generator and if it takes additionally arguments such as y or z, it automagically supplies the values to it at runtime